"Predict. Prevent. Rebuild Lives."

ReformAI is an AI-powered Android application designed to help law enforcement, NGOs, and policymakers predict recidivism risk (likelihood of reoffending) and recommend rehabilitation programs. It also includes a Safety by Area Index, which allows users to check the safety levels of different locations based on crime data.

• 📊 Predicts the risk level (1-10) of a released offender reoffending.
• 🛠 Uses crime history, employment status, drug use, and gang affiliation to assess risk.
• 📢 Suggests rehabilitation programs (Job placement, mental health therapy, or intensive rehab).

• 🌍 Displays safety levels for different areas using a color-coded system:
  • 🟢 Safe Area (Green)
  • 🟠 Unsafe Area (Orange)
  • 🔴 Very Unsafe Area (Red)
• 🔍 Uses real-time crime data (FBI API, Kaggle datasets, local police reports).
• 📍 Users can enter a location to check crime density & severity.

• 🗂️ Stores past risk assessments & safety index checks.
• 📊 Allows law enforcement & NGOs to track high-risk individuals.

• Language: Java & XML
• Networking: Retrofit for API calls
• UI Components: RecyclerView, CardView, Google Maps API

• Machine Learning: Scikit-Learn (Random Forest/XGBoost)
• Database: PostgreSQL/MongoDB (For case history tracking)
• Crime Data Sources: FBI API, Kaggle, OpenCrimeData

1. User enters offender details (age, crime type, prior convictions, employment status, etc.).
2. ReformAI sends data to Flask API.
3. ML model calculates risk score (1-10).
4. ReformAI suggests rehabilitation programs.

1. User enters a location (city, zip code).
2. ReformAI fetches crime data from APIs (FBI, Kaggle).
3. ReformAI calculates safety score:

   Safety Score = 100 - (w1 × Crime Rate + w2 × Criminal Density + w3 × Crime Severity)
   

4. ReformAI displays safety level (Green, Orange, Red).

✅ Google Maps Integration → Display crime heatmaps.
✅ Real-time Notifications → Alert users if an area becomes unsafe.
✅ Bias Mitigation → Ensure fairness in recidivism predictions.

• Mahima Lolla
• Siddharth Shukla
• Sahil Satramani
• Ujjwal Pandit
• Suresh Palaparthi

For any issues or contributions, please open a GitHub issue or email [email protected].